In many real shallow water flow phenomena particularly in near-field situations – such as flows around structures, outlets, junctions, etc. – the effects of the horizontal stresses become significant on the water, leading to a turbulent condition. Common models based on the shallow water equations assume that the water is inviscid, thus automatically neglecting the effects of such stresses. Even though some viscous shallow water-based models have also been developed, they still cannot represent the turbulence effects, since the diffusive fluxes in those models distinguish the effects of the turbulent fluxes. Therefore, we extended in this thesis two turbulence models – the depth-averaged parabolic eddy viscosity model and the depth-averaged mixing length model – into the 2D shallow water equations leading to the 2D depth-averaged Reynolds-averaged Navier-Stokes equations provided within the framework of an in-house code called NUFSAW2D. The code was parallelized using OpenMP. A parallel efficiency of approximately 92.2% (weak scaling) and an average speedup of 6.5x (strong scaling) were achieved for more than 1 million cells (2 million edges) using 16 cores, in which the results obtained using 2 cores were used as the benchmark. These indicated a good scalability result for large scale applications and would be suitable for a more advanced development, i.e. hybrid parallelization technique combining both OpenMP and MPI in the future.     «

In many real shallow water flow phenomena particularly in near-field situations – such as flows around structures, outlets, junctions, etc. – the effects of the horizontal stresses become significant on the water, leading to a turbulent condition. Common models based on the shallow water equations assume that the water is inviscid, thus automatically neglecting the effects of such stresses. Even though some viscous shallow water-based models have also been developed, they still cannot represent...     »